Thin-film composite forward osmosis membranes functionalized 1 with graphene oxide − silver nanocomposites for biofouling control

Innovative approaches to prevent bacterial attachment and biofilm growth on membranes are critically needed to avoid decreasing membrane performance due to biofouling. In this study, we propose the fabrication of anti-biofouling thin-film composite membranes functionalized with graphene oxide-silver nanocomposites. In our membrane modification strategy, carboxyl groups on the graphene oxide-silver nanosheets are covalently bonded to carboxyl groups on the surface of thin-film composite membranes via a crosslinking reaction. Further characterization, such as scanning electron microscopy and Raman spectroscopy, revealed the immobilization of graphene oxide-silver nanocomposites on the membrane surface. Graphene oxide-silver modified membranes exhibited an 80% inactivation rate against attached . Pseudomonas aeruginosa cells. In addition to a static antimicrobial assay, our study also provided insights on the anti-biofouling property of forward osmosis membranes during dynamic operation in a cross-flow test cell. Functionalization with graphene oxide-silver nanocomposites resulted in a promising anti-biofouling property without sacrificing the membrane intrinsic transport properties. Our results demonstrated that the use of graphene oxide-silver nanocomposites is a feasible and attractive approach for the development of antibiofouling thin-film composite membranes. Disciplines Engineering | Science and Technology Studies Publication Details Faria, A. F., Liu, C., Xie, M., Perreault, F., Nghiem, L. D., Ma, J. & Elimelech, M. (2017). Thin-film composite forward osmosis membranes functionalized with graphene oxide-silver nanocomposites for biofouling control. Journal of Membrane Science, 525 146-156. Authors Andreia Faria, Caihong Liu, Ming Xie, Francois Perreault, Long D. Nghiem, Jun Ma, and Menachem Elimelech This journal article is available at Research Online: http://ro.uow.edu.au/eispapers/6294


53
Global demand for drinking water is expected to increase in the coming decades due to 54 rapid population growth and climate change [1]. Membrane-based water purification 55 processes play a crucial role in mitigating water scarcity worldwide [1,2]. Due to their high 56 permeate water flux and salt rejection capabilities, thin-film composite (TFC) membranes 57 have been considered the state-of-the art for water desalination technologies such as reverse 58 osmosis (RO) and forward osmosis (FO) [1][2][3][4]. Despite these advantages, TFC membranes 59 encounter several operational limitations. One significant challenge is the attachment of 60 microorganisms and subsequent biofilm formation [5,6]. 61 The growth of bacteria as biofilms can affect membrane performance by decreasing 62 permeate water flux and salt rejection [6]. Furthermore, biofouling development can lead to 63 an increase in energy consumption [5][6][7]. Ordinary procedures such as pretreatment and 64 chemical cleaning are being used to mitigate biofouling [5,6]. However, no pre-treatment can 65 completely eliminate biofouling, and it is well known that the polyamide layer of TFC 66 membranes undergoes degradation in the presence of chemical oxidants such as chlorine [8]. 67 Therefore, there is a critical need to develop innovative strategies to control microbial 68 proliferation at the membrane surface. 69 Several studies have proposed to modify the surface of TFC membranes with polymers 70 [9], bio-active molecules [10], or antimicrobial nanomaterials [11] in order to impart 71 antimicrobial activity and biofouling resistance to the membrane. For instance, it has been 72 shown that TFC membranes functionalized with silver or copper nanoparticles presented a 73 diminished susceptibility to biofouling [12,13]. Alternatively, carbon-based nanomaterials 74 such as carbon nanotubes (CNTs) and graphene oxide (GO) have also been linked to the 75 polyamide layer to generate TFC membranes with enhanced antimicrobial properties [14][15][16]. 76 Antimicrobial nanomaterials can be incorporated by embedding them within the 77 membrane polymeric matrix [17]. Post-fabrication modification, on the other hand, is focused 78 on the immobilization of nanomaterials at the membrane surface via physical interactions 79 [13], chemical binding [16], or layer-by-layer techniques [18]. Because the nanomaterials are 80 placed specifically at the membrane surface, post-fabrication functionalization is unlikely to 81 affect significantly the properties of the polyamide layer [15,16]. This technique is also 82 material-and cost-efficient since fewer nanomaterials are required to tailor the membrane 83 surface chemistry [15,16]. 84 water and additionally purified by dialysis (3, 500 Da membranes, Spectrum Laboratories,  150 Inc., CA, USA) for three or four days. The final brown suspension was frozen in liquid 151 nitrogen, dried by lyophilization, and stored at room temperature. 152 GOAg nanocomposites were synthesized by employing Tollens' modified method, 153 which is based on the complexation of Ag + ions with NH 4 OH and further reduction using 154 saccharides [29]. To prepare GOAg nanocomposites, GO (12.5 mg) was dispersed in DI 155 water (35 mL) and bath-sonicated (Aquasonic Model 150T) for 30 minutes. AgNO 3 (8.65 156 mg) was dissolved in 5 mL of DI water and then combined with a 50 mM NH 4 OH solution (5 157 mL). The resulting solution was stirred for 10 minutes. Then, the silver solution was 158 introduced to the prior GO dispersion and the mixture was bath-sonicated for an additional 20 159 minutes. Immediately after sonication, 5 mL of a glucose solution (100 mM) was added by 160 drops. The reaction was conducted overnight at room temperature. After synthesis 161 completion, the color of the suspension changed from brown to green-blue, indicating the 162 nanocomposite formation. To remove the excess of chemical residues, the GOAg 163 nanocomposite suspension was purified by dialysis (3,

Functionalization of TFC membranes with GO and GOAg nanocomposites 173
The polyamide active layer of thin-film composite (TFC) membranes was 174 functionalized with GO or GOAg using a well-stablished method adapted from previous 175 studies [15,16]. Pristine TFC membranes were placed in frames and sealed with clips to 176 avoid any leakage. With only the active (top) surface exposed, the membranes were kept on 177 an orbital shaker at 60 rpm at room temperature throughout the functionalization procedure. 178 GO and GOAg nanocomposites were chemically bound to the TFC membranes using 179 EDC and NHS as crosslinks. The entire functionalization process can be divided into three 180 steps. The first step is the activation of the native carboxylic functional groups on TFC 181 membranes. For this, EDC (4.0 mM) and NHS (10.0 mM) were dissolved in 10 mM MES 182 buffer (pH 5.0) and left to react with the membrane surface for two hours. Next, the solution 183 was removed and the membrane surface was rinsed twice with DI water. In the presence of 184 EDC and NHS, the native carboxyl functionalities on the membrane surface were converted 185 to reactive ester groups. In the second step, the activated carboxyl groups were reacted with 186 ethylene diamine (ED) (10 mM) in a 0.15 M NaCl and 10 mM HEPES buffer (pH 7.5) for 187 one hour to yield an amine-terminated membrane surface. The membrane surface was then 188 rinsed twice with DI to remove unbound ED. 189 The third step comprises the activation of the carboxylate groups on GO and GOAg by 190 EDC and NHS, as described earlier for the pristine TFC membrane. Twenty-five milliliters of 191 the GO and GOAg dispersions (250 µg mL -1 ) were diluted with 20 mL of 10 mM MES buffer 192 (pH 6). EDC (1.5 mM) and NHS (2.5 mM) were dissolved in 5 mL of MES buffer (pH 6.1) 193 and slowly poured into the GO and GOAg dispersions. The system was kept stirring for 30 194 minutes at room temperature. EDC and NHS decreased the buffer pH to 5.  contact with the membrane surface, pH was adjusted to 7.2 using a sodium hydroxide 196 solution (1 M). After activation, the ED-functionalized membrane coupons were brought into 197 contact with 20 mL of the activated GO and GOAg samples, and the system was gently 198 stirred at room temperature for three hours. The intermediate reactive esters on GO and 199 GOAg react with the primary amine functional groups, thus irreversibly binding the 200 nanomaterials to the membrane surface. At the end of the reaction, the membranes were 201 rinsed to wash out the unbound materials and restore the unreacted carboxyl groups. The TFC 202 membranes modified with GO or GOAg are referred to as TFC-GO and TFC-GOAg, 203 respectively. 204

Membrane characterization 205
The presence of GO and GOAg nanocomposites on the membrane surface was 206 confirmed by scanning electron microscopy (SEM) using an XL-Philips scanning electron 207 microscope. A Cressington (208 carbon) sputtering machine was applied to coat the sample 208 with a thin layer (10-20 nm) of carbon. Images were taken at an acceleration voltage of 10 209 kV. Energy dispersive spectroscopy (EDS) was utilized to detect the presence of silver. The transport properties of the membrane were determined in a cross-flow FO filtration 226 system according to a four-step method reported in our previous publication [30]. Briefly were also determined. 291 Biofilm total organic carbon (TOC) and protein concentration were also quantified. 292 For TOC measurements, membrane sub-sections (2 cm × 2 cm) were re-suspended in 24 mL 293 sterile wastewater with 10 µL of 1 M HCl. Samples were then sonicated on ice in three  second cycles to remove organic content from the membrane. TOC in the resultant solution 295 was then analyzed using a TOC analyzer (TOC-V, Shimadzu, Japan). TOC concentrations 296 were normalized by membrane sample area. For protein quantification, membrane sub-297 sections (2 cm × 2 cm) were cut and suspended in 2 mL Eppendorf tubes with 1 mL 1X 298 Lauber buffer (50 mM HEPES (pH 7.3), 100 mM NaCl, 10% sucrose, 0.1% CHAPS, and 10 299 mM DTT) and probe sonicated on ice (three 30-second cycles) using an ultra-cell disruptor. 300 The membrane was then removed and cell extracts were centrifuged at 12,000 rpm for 10 301 minutes to remove detritus matter. The supernatant was then collected for protein 302 quantification using a BCA protein assay kit (Thermo Scientific, IL). 303

Physicochemical characteristics of GO and GOAg nanocomposites 305
The chemical exfoliation of graphite produces a brown dispersion composed of single-306 layer graphene oxide (GO) sheets ( Figure 1A). A typical GO sample characteristically has a 307 wide size distribution. GO average size is dependent on several factors such as time of 308 reaction, the graphite precursor, and the concentration and type of oxidizing agent used 309 during sample preparation. The SEM image of an aqueous suspension of our prepared GO 310 (Figure 2) shows the presence of flat sheets with an average area of 0.36 ± 0.37 µm 2 . The 311 average size of GO sheets has been shown to influence their reactivity, in particular the 312 cytotoxicity to bacterial cells [28,32]. For the preparation of GOAg nanocomposites, GO powder was dispersed in DI water 324 and mixed with the precursor AgNO 3 . The reaction was conducted at alkaline conditions due 325 to the addition of ammonium hydroxide (NH 4 OH); glucose (dextrose) was used as a reducing 326 agent. The change in color from brown to green-blue was an indicator of the decoration of 327 GO sheets with AgNPs ( Figure 1A). Previous studies have reported the use of sugar to 328 reduce Ag + ions to silver nanoparticles [29]. This method is widely known as the Tollens 329 reaction. The mechanism involves the interaction of Ag + ions with NH 4 OH to form 330 intermediate species (Ag(NH 3 ) 2 ) + that are then reduced to nanoclusters upon contact with the 331 sugar molecules [33]. It is worth mentioning that the reducing property of monosaccharides, 332 such as glucose, is attributable to the presence of free aldehyde or ketone functional groups 333 on the sugar molecules. In comparison to many of the processes already reported in the 334 literature, the Tollens method is advantageous since it applies a non-toxic and an 335 environmental friendly molecule as a reducing agent. Moreover, the chemical reaction does 336 not require high temperatures or the use of aggressive organic solvents [23,24,34,35].  311 crystalline planes of AgNPs, respectively [36]. Thermogravimetric analysis was carried 347 out to investigate the thermal decomposition pattern of both GO and GOAg ( Figure S1C). 348 TGA curves also provide information about the silver content in the GOAg sample [23,24]. 349 The residues above 600°C indicate that the relative content of silver is approximately 10 wt 350 % of the total GOAg nanocomposites ( Figure S1C). The decoration of GO sheets with AgNPs was confirmed by transmission electron 358 microscopy (TEM), as shown in Figure 1C. The AgNPs appeared as black dots distributed 359 throughout the graphene surface with an average size of 16 ± 12 nm ( Figure 1D). Both GO 360 and GOAg nanocomposites showed a wrinkled and paper-like morphology on the TEM 361 images (Figures 1B and C). Since particles were not found detached from GO sheets, we 362 surmise the nucleation occurs preferentially on the graphene surface. The negatively charged 363 oxygen-containing functional groups on GO likely offer nucleation sites for the Ag + ions via 364 electrostatic interaction [23,35]. Once adsorbed on GO sheets, Ag + ions can be reduced to 365 Ag 0 nanoparticles in the presence of a reducing agent. The physicochemical characteristics of 366 GOAg nanocomposites may differ depending on the degree of oxidation of GO sheets and the 367 initial concentration of silver utilized [37,38]. 368 369 370

GO and GOAg sheets are covalently bound to the membrane surface 371
The binding of GO and GOAg nanocomposites to TFC membranes was developed 372  Figure 4A) [13,15,16]. The areas where the polyamide layer 393 was modified with GO or GOAg sheets appeared as dark spots on the membrane surface 394 (Figures 4B and C). No such dark spots are present on the SEM images of pristine 395 membranes ( Figure 4A). The rough surface of the polyamide layer seems to be covered by 396 GO or GOAg nanosheets and small bright features (~50 nm) were detected on the surface of 397 TFC-GOAg membranes (Figure 4C). Energy dispersive spectroscopy (EDS) spectrum 398 acquired directly from those bright spots revealed a peak at 4.0 keV that is attributable to 399 silver ( Figure 4D). A visual inspection indicated that TFC membranes did not present drastic 400 changes in color after binding GO or GOAg nanocomposites. In addition to SEM imaging, GO and GOAg-modified membranes were characterized 409 by Raman spectroscopy (Figure 5). The functionalization of TFC membranes with both 410 nanomaterials was indirectly confirmed through changes in the intensity ratio between the 411 peaks at 1148 and 1620 cm -1 (I 1148 /I 1620 ), as reported in our previous publication [16]. Among 412 several absorption peaks, Raman spectrum of TFC membranes is particularly characterized 413 by the presence of symmetric C-O-C stretching (~1148 cm -1 ) and phenyl ring vibration 414 (~1590-1620 cm -1 ) [39]. It is already well known that bare GO displays two reference peaks 415 at 1350 cm -1 (D band) and 1590 cm -1 (G band) in the Raman spectrum [40]. With the binding 416 of GO and GOAg nanocomposites, the intensity of the peak at 1148 cm -1 is expected to 417 decrease, whereas the intensity of the peak at 1620 cm -1 is likely to increase due to the 418 contribution of the G band from GO sheets. Changes in surface hydrophilicity were investigated through static water contact angle 454 measurements ( Figure S2). However, no significant differences in contact angle were noticed 455 after functionalization of TFC membranes with either GO (32.6 ± 2.8°) or GOAg sheets (33.8 456 ± 6.2°), despite the large amount of hydrophilic, oxygen-containing functional groups on the 457 graphene sheets. One reason for this observation is the already relatively very low contact 458 angle of the pristine TFC membranes (38.1 ± 1.9°). 459

Functionalization with GOAg nanocomposites does not affect membrane transport 460
properties 461 One of the greatest challenges of modifying the surface of TFC membranes is to ensure 462 that water permeability (A) and salt selectivity (B) are not affected by the binding of 463 polymeric molecules or nanomaterials. Figure 7B summarizes the A, B, and S parameters for 464 TFC, TFC-GO, and TFC-GOAg membranes. We observed that the A and B coefficients did 465 not significantly change with the binding of GO or GOAg to the membrane surface (p > 466 0.05), even though TFC-GOAg presented a small decrease in the water permeability 467 coefficient A compared to the unmodified membrane ( Figure 7B). The salt permeability 468 coefficient B slightly increased from 1.33 ± 0.21 L m -2 h -1 for the pristine membrane to 1.64 ± 469 0.32 and 1.59 ± 0.21 L m -2 h -1 for TFC-GO and TFC-GOAg membranes, respectively. As 470 expected, Figure 7B also reveals that the membrane structural parameter S of the pristine 471 TFC membrane did not change by our modification procedure. These results indicate that the 472 functionalization with GO or GOAg does not impact the transport properties of the 473 membrane polyamide layer. This result is consistent with our previous work, where the 474 modification of RO TFC membranes with pristine GO did not change the membrane 475 transport properties [16,41]. Similar observations have also been reported for TFC RO 476 membranes modified with multiple layers of GO sheets [42]. This low impact of GO on the 477 membrane performance is probably due to its atomic thickness and hydrophilic nature. Table  478 S1 presents one full set of experimental data (measured water and reverse salt fluxes and 479 relevant coefficients of determination, R 2 ) for the TFC, TFC-GO, and TFC-GOAg

Bacterial attachment and viability are significantly suppressed by GOAg 498
Antimicrobial activity was first evaluated after exposing the membrane surface to P. 499 aeruginosa cells for three hours. In comparison to pristine TFC, the TFC-GO membrane 500 displayed no toxic effect towards P. aeruginosa ( Figure 8A). TFC-GOAg membrane, on the 501 other hand, exhibited a bacterial inactivation rate of around 80% against P. aeruginosa, 502 relative to the non-modified TFC membranes. In other words, the number of viable cells on 503 TFC-GOAg was significantly lower than that of the unmodified control, implying that 504 functionalization with GOAg imparted a strong antimicrobial activity to the membrane 505 surface. 506 AgNPs, and the mechanism of toxicity can be explained by both release of toxic Ag + ions and 523 direct contact with the AgNPs on the membrane surface [25,43]. The high affinity of silver 524 for thiol (-SH) functional groups of proteins may damage the stability and architecture of the 525 bacterial cell wall through the generation of holes and vacancies [44,45]. Disruption of cell 526 wall structure could irreversibly affect the transport of nutrients, thus inactivating the 527 bacterial cells. 528 529

GOAg nanocomposite functionalized membranes exhibit reduced biofouling rate. 530
The anti-biofouling properties of TCF and TFC-GOAg membranes were investigated 531 by allowing P. aeruginosa cells to grow on the membrane surface for 24 hours in a dynamic 532 cross-flow biofouling test. One of the consequences of biofilm formation on TFC membranes 533 is the decrease in permeate water flux. As shown in Figure 9A, the development of biofilm 534 on pristine TFC membrane resulted in a flux decline of approximately 50%. However, when 535 TFC membrane is functionalized with GOAg nanocomposites, the flux decline is 536 significantly reduced. The difference in the water flux behavior is attributable to differences 537 in the structure and composition of the biofilms on the pristine TFC and TFC-GOAg 538

membranes. 539
To obtain information about the biofilm properties, the biofouled membranes were 540 characterized by confocal microscopy.  Figure 9D) than on TFC-GO or pristine TFC membranes (Figures 9B and C). The 544 dead cell region reached the top layer of the biofilm on the TFC-GOAg membrane, indicating 545 that direct contact with the GOAg nanocomposite was not required and that silver ions could 546 leach and diffuse to the upper cell layers. Therefore, the addition of Ag in a GOAg 547 nanocomposite played a key role in mitigating biofilm development on TFC membranes. 548 GOAg membranes. The biofilm was grown after 24-hour biofouling runs as described in (A). 559 Live cells, dead cells, and exopolysaccharides were stained with Syto 9 (green), propidium 560 iodide (red), and Con A (blue) dyes, respectively. 561 562 Table 1 summarizes the biofilm properties for TFC, TFC-GO, and TFC-GOAg  563 membranes. For instance, the biofilm on TFC-GOAg membrane was almost two times 564 thinner than that on the pristine TFC membrane. Furthermore, the live cell biovolume (µm 3 565 µm -2 ) on TFC-GOAg was decreased by almost 50% compared to the non-modified TFC 566 membrane. The lack of antimicrobial activity for TFC-GO membrane is explained by the 567 relatively large size of the GO sheets [28]. The results observed from confocal imaging are in 568 accordance with the CFU counts reported in Figure 8A, where pristine TFC and TFC-GO 569 membranes exhibited similar antimicrobial properties. The biofilm contents of protein and 570 total carbon were also drastically reduced after modification of TFC membranes with GOAg 571 nanocomposites. The total protein mass was diminished from 18.7 ± 2.5 to 9.1 ± 6.2 pg µm -2 572 after binding GOAg sheets to the membrane surface ( Table 1). 573 Our findings suggest that bacterial growth on the TFC membrane surface was strongly 574 inhibited by GOAg nanocomposites. The decrease in the number of live cells on the TFC-575 GOAg membrane led to a significant reduction in biofilm thickness, live cell biovolume, and 576 EPS production (

Conclusion
In this study, we report the synthesis of GOAg nanocomposites and their further application as antimicrobial agents for the control of biofouling in forward osmosis membranes. GOAg nanocomposites were prepared through a straightforward process whereby silver nanoparticles are in-situ nucleated on GO sheets. The formation of silver nanoparticles on GO sheets is done by using glucose as a reducing agent. The resulting GOAg nanocomposites displayed silver nanoparticles with an average size of 16 nm which were bound irreversibly on the GO surface. Carboxylic groups on GOAg were used as target points to bind the graphene sheets to the amine-terminated polyamide layer. The surface modification of TFC membranes with GO or GOAg nanocomposites was successfully demonstrated by SEM and Raman spectroscopy analyses. We also show that the intrinsic transport properties of TFC membranes were not affected by the modification with GO or GOAg nanocomposites.
Static antimicrobial assays showed that GOAg modified membranes were able to significantly inhibit the attachment of Pseudomonas aeruginosa cells. Unlike some previous studies, the membrane modified just with GO showed no toxicity to bacterial cells. In addition, dynamic biofouling experiments performed using a bench-scale FO system demonstrated the anti-biofouling property of membranes modified with GOAg sheets. A massive amount of dead cells can be seen on the confocal images taken from TFC-GOAg membranes. In addition, the biovolume of live cells was substantially decreased for membranes modified with GOAg. Dynamic biofouling experiments also showed that the flux decline due to biofouling development was reduced by 30% after modification of TFC membranes with GOAg nanocomposites. Our results suggest that membrane functionalization with GOAg is a robust platform to yield TFC membranes possessing enhanced biofouling resistance.   Table S1: Estimation of water and salt permeability coefficients of TFC, TFC-GO, and TFC-GOAg membranes by the FO four-step characterization method [1]. The final water permeability coefficient A, salt permeability coefficient B, and structural parameter S presented in the manuscript were determined from three sets of independent measurements for each membrane.